To the Editors:

We appreciate the comments made by Drs. Mariette, Reynolds, and Emery regarding our review. We do not agree with the take-home message from their letter. The authors state that the results of studies on anti-TNF agents and cancer risk in patients with autoimmune disease and cancer are remarkably consistent, and they suggest that while patients with autoimmune disease are at increased risk for cancer as compared with the general population, this risk is not further increased by the use of anti-TNF agents.

By focusing largely on lymphoma data from one registry, the Swedish ARTIS registry, the authors only underscore our point; there are considerable conflicting data about the contribution of anti-TNF agents to cancer risk that are inadequately addressed by studies performed to date. It bears repeating that there are considerable data to indicate that the general malignancy risk for patients with autoimmune diseases receiving anti-TNF therapy is higher than that for patients matched for disease but not treated with anti-TNF agents. As one example, a trial of infliximab for chronic obstructive pulmonary disease (COPD) treatment showed no benefit for COPD, but it did show an increase in solid organ malignancies in patients treated with infliximab over the first year of treatment as compared with COPD patients who did not receive the drug (1). In another example, Chakravarty et al showed an increase in nonmelanoma skin cancer in patients with RA treated with TNF inhibitors as compared with RA patients treated with other agents (2). This association with nonmelanoma skin cancers and anti-TNF therapy was also seen in a second larger US observational study (3). As a third example, we refer to the clustering of a rare lymphoma subtype in children with inflammatory bowel disease treated with monoclonal antibodies to TNF (4, 5).

Based upon these and considerable other data, we argue that it is difficult to conclude that the excess risk for cancer in patients with autoimmune diseases treated with anti-TNF agents is due solely to their underlying disease. We believe that the definitive prospective clinical trial testing the association between cancer and anti-TNF therapy is yet to be performed. In our review, after discussing the data suggesting an association with cancer, we then endeavor to present the complexities inherent in undertaking this sort of study.

As well as clinical trials, a better understanding of the molecular mechanisms of autoimmune diseases and their therapies remains an extremely important component of the study of cancer risk in patients with autoimmune disease and identifying those patients at greatest risk. Studies like those by Demirkaya et al (6), in which they find that lymphocytes from children with JIA treated with anti-TNF agents were more sensitive to genotoxic stress and less efficient at DNA repair than were cells from the same patients prior to the initiation of anti-TNF therapy, may point to important pathways and potential tests for clinical use. If a subset of patients with JIA or RA exhibit increased spontaneous levels of DNA damage or increased genotoxicity when treated with anti-TNF agents, for example, then these patients may represent a subset of susceptible patients who would benefit from close followup to detect the development of malignancies, or who should be treated with other drugs not associated with malignancy, such as agents that block T cell costimulation. In the French RATIO registry, although the risk of lymphoma was similar for those treated and not treated with anti-TNF agents, there were lymphomas associated with immunosuppression, and the risk of lymphoma varied with the medication used (7). Identification of an at-risk population that should be treated differently or monitored more intensively is critical. Additionally, these studies may suggest intermediate phenotypes associated with cancer risk. These can then be used in future pharmacogenetic studies to dissect the genetic basis for interindividual differences in DNA damage that may ultimately yield clinical tests to predict those in whom the drugs would most benefit and those in whom the risks are substantial. Indeed, Dr. Mariette made this very point in a prior manuscript when he pointed to the importance of “studies combining genotypic and phenotypic information about the autoimmune disease and its treatment, with detailed data on lymphoma occurrence with the ultimate aim of identifying heterogeneities in risk rather than the average” (8). We would agree that these studies are critical.

In conclusion, TNF blockade has been a remarkable breakthrough in the treatment of a wide variety of inflammatory diseases and has led to an improved quality of life for millions of people around the world. Nothing is without a cost, however. We do not believe that the data indicate that anti-TNF therapy is not associated with an increased risk of malignancy. Rather, we believe that this controversial issue has not yet been adequately examined for the reasons we discuss in our review. Continued study with careful phenotypic and genotypic evaluation of autoimmune patients with and without cancer is required before conclusions can be drawn. As a result, in summary, the message for patients should be that the jury is still out.

Karen Onel MD*, Kenan Onel MD, PhD*, * University of Chicago, Chicago, IL.